rados/osd: enable compact_running perfcounter at PRIO=5

The compact_running perfcounter is a bool indicator for rocksdb compaction. By enabling this counter at a USEFUL priority level,
performance monitoring can indicate whether compaction is a factor.

Fixes:  https://tracker.ceph.com/issues/73591
Signed-off-by: Paul Cuzner <pcuzner@ibm.com>

diff --git a/src/kv/RocksDBStore.cc b/src/kv/RocksDBStore.cc
index 0c1d4c74192b31278945acd15eafe4b63615b1a1..f0351c4ad8e520564dc994bc1b1592233923fdfd 100644 (file)
--- a/src/kv/RocksDBStore.cc
+++ b/src/kv/RocksDBStore.cc
@@ -1254,7 +1254,7 @@ int RocksDBStore::do_open(ostream &out,
   plb.add_time_avg(l_rocksdb_submit_latency, "submit_latency", "Submit Latency");
   plb.add_time_avg(l_rocksdb_submit_sync_latency, "submit_sync_latency", "Submit Sync Latency");
   plb.add_u64_counter(l_rocksdb_compact, "compact", "Compactions");
-  plb.add_u64_counter(l_rocksdb_compact_running, "compact_running", "Running compactions");
+  plb.add_u64_counter(l_rocksdb_compact_running, "compact_running", "Running compactions", nullptr, PerfCountersBuilder::PRIO_USEFUL);
   plb.add_u64_counter(l_rocksdb_compact_completed, "compact_completed", "Completed compactions");
   plb.add_time(l_rocksdb_compact_lasted, "compact_lasted", "Last completed compaction duration");
   plb.add_u64_counter(l_rocksdb_compact_queue_merge, "compact_queue_merge", "Mergings of ranges in compaction queue");